Automatic generation of decomposition based matrix inversion architectures

Abstract

Matrix inversion is an essential computation for various algorithms which are employed in multi-antenna wireless communication systems. FPGAs are ideal platforms for wireless communication; however, the need for vast amounts of customization throughout the design process of a matrix inversion core can overwhelm the designer. Decomposition methods provide the analytic simplicity and computational convenience necessary for computationally intensive matrix inversion. This paper presents automatic generation of different decomposition based matrix inversion architectures using a matrix inversion core generator tool, GUSTO with different parameterization options. We present automatic generation of a variety of general purpose matrix inversion architectures which have comparable results to published matrix inversion architecture implementations, but offer the advantage of providing the designer the ability to study the tradeoffs between architectures with different design parameters.